# AI tutors could narrow education gaps in poorer countries

Markdown copy of https://realalpha.app/s/ai-tutors-could-narrow-education-gaps-in-poorer-countries/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

- Speaker: [Reed Hastings](https://realalpha.app/people/reed-hastings/) · Reed Hastings is cofounder and former CEO of Netflix. · Co-founded Netflix and led it as CEO, then became majority owner and chair of Powder Mountain.
- Host: [Reid Hoffman](https://realalpha.app/people/reid-hoffman/) · Co-founder, Manas AI · Co-founded LinkedIn; one of OpenAI's first investors; Greylock partner.
- Show: Possible
- Episode: [Reed Hastings: From Building Netflix to What Comes Next](https://realalpha.app/v/2qoGbd9d2yA/), Apr 22, 2026
- Moment: 39:02 to 41:32 of the episode (3 min), playing on https://realalpha.app/s/ai-tutors-could-narrow-education-gaps-in-poorer-countries/; the episode from there: https://realalpha.app/v/2qoGbd9d2yA/?t=2342

## Quote

> “the education budget might be $**300 per kid per year**. And the class size might be 50 to 70 kids.”
> — [Reed Hastings](https://realalpha.app/people/reed-hastings/) at [39:19](https://realalpha.app/v/2qoGbd9d2yA/?t=2359) · Possible

## Takeaways

1. Some schools have $300 per student per year and classes of 50 to 70.
2. AI tutors could reach students where teacher supply is limited.
3. Hastings says earlier laptop efforts were about 20 years too early.

## About this segment

Reed Hastings sees AI tutors, tablets and school connectivity as ways to expand learning in lower-income countries. He argues that cheap phones make the approach more viable than earlier laptop schemes.

From “Reed Hastings: From Building Netflix to What Comes Next”, Apr 22, 2026. [Watch the full episode](https://realalpha.app/v/2qoGbd9d2yA/)

## Transcript

Transcribed by machine (Whisper) from the episode’s audio, so a word may be off here and there.

[39:07](https://realalpha.app/v/2qoGbd9d2yA/?t=2347) education going to change outside the United States with AI? Well, I think what we'll see is tremendous flourishing of AI teacher in an international situation. So we have pretty good teachers in the U.S. and pretty high satisfaction and rich countries generally here. But in lower income countries, the education budget might be $300 per kid per year. And the class size might be 50 to 70 kids. And there's just not a lot of learning going on.

[39:33](https://realalpha.app/v/2qoGbd9d2yA/?t=2373) And I think the combination of a Starlink on every school and tablet for every child and then really good AI software will close the gap. So think of the way mobile phones are ubiquitous in the developing world. I think AI learning will be ubiquitous. And that will help. I don't know that they'll leapfrog the advanced countries, but they'll certainly close the gap that we've seen historically and in places.

[40:00](https://realalpha.app/v/2qoGbd9d2yA/?t=2400) They may leapfrog. I feel like lately, there's been like breathless articles everywhere about how one laptop per child was the hope. And then it was really a failure. And we put so much money into it, and it didn't work. And then other people are saying like, well, that's really nice that you want that. But you know, the cost of compute is too much. Like, what would you say to those people who think that like, we've heard tech solutions and education before, but it's not going to happen this time? Yeah, there's always a crash dream phenomena. Like, you know, in the 1980s, I was in the AI wave that was expert systems and they were public companies and fifth generation and the Japanese were going to be this big threat and you know and that AI didn't work but this one is so just think just because it failed before it was too early it doesn't mean it won't work so

[40:45](https://realalpha.app/v/2qoGbd9d2yA/?t=2445) I think the one tablet per child around the world is very scalable built on the mobile phone platforms and it is some of the same ideas that one laptop per child had but that was just 20 years too early. Yep. And I assume you think the cost of compute is just going to go through the floor and so that it won't be a barrier to entry for folks in the developing world. That's right. The costs are really driven off the phone market. So, you know, we have $800 phones in the U.S., but in most of the world, you know, there's $50 phones are quite usable. And so think Think of it as a $50 phone and a Starlink per school and some solar panels, you know.

[41:27](https://realalpha.app/v/2qoGbd9d2yA/?t=2487) So it's very scalable. Fantastic.